Orbifold stability and Miyaoka-Yau inequality for minimal pairs (Q2105809)

A complex-projective manifold \(X\) with \(K_X\) ample famously admits a Kähler-Einstein metric. As a consequence, we have the Miyaoka-Yau inequality \[ \big( 2(n+1)\mathrm{c}_2(X) - n\mathrm{c}_1^2(X) \big) \cdot K_X^{n-2} \ge 0, \] where \(n = \dim X\). Theorem A of the present paper extends this inequality to minimal models, i.e.~normal complex-projective varieties with terminal singularities such that \(K_X\) is \(\mathbb{Q}\)-Cartier and nef. Of course, the statement is empty if e.g. \(K_X\) is numerically trivial; Theorem B proves the following more general version: \[ \big( 2(n+1)\mathrm{c}_2(X) - n\mathrm{c}_1^2(X) \big) \cdot K_X^i \cdot H^j \ge 0, \] where \(H\) is ample and \(i = \mathrm{min}(\nu, n-2)\), where \(\nu\) is the numerical dimension of \(K_X\). Actually, Theorem B deals even more generally with dlt pairs \((X, D)\) with standard coefficients. (This generality is also needed in the proof.) The development of the necessary orbifold techniques is one of the main technical contributions of the paper. The other is Theorem C, which proves the semistability of the (orbifold) tangent sheaf in the above setting. The proof of the latter is analytic, relying on the theory of conical/cuspidal metrics.